A New Series of Cobalt and Iron Clathrochelates with Perfluorinated Ribbed Substituents.

The study tackles one of the challenges in developing platinum-free molecular electrocatalysts for hydrogen evolution, which is to seek for new possibilities to ensure large turnover numbers by stabilizing electrocatalytic intermediates. These species are often much more reactive than the initial electrocatalysts, and if not properly stabilized by a suitable choice of functionalizing substituents, they have a limited long-time activity. Here, we describe new iron and cobalt(II) cage complexes (clathrochelates) that in contrast to many previously reported complexes of this type do not act as electrocatalysts for hydrogen evolution. We argue that the most probable reason for this behavior is an excessive stabilization of the metal(I) species by perfluoroaryl ribbed groups, resulting in an unprecedented long-term stability of the metal(I) complexes even in acidic solutions.

Introduction

n class="Chemical">Cobaltn> clathrochelates,[1] which are polyazomethine-based cage complexes, have been recently recognized as efficient electrocatalysts for the hydrogen evolution reaction (HER) at low overpotentials;[2] however, the mechanism of their electrocatalytic activity is still not fully understood. In some cases, cobalt(II) clathrochelates were shown to function only as precatalysts that produced electrocatalytically active nanoparticles[3] under acidic conditions, whereas in others, they themselves were homogenous electrocatalysts that required a very small overpotential.[2c] In contrast to the former, the latter have halogen atoms as their ribbed substituents; they keep the potential of the Co2+/+ reduction close to the thermodynamic potential for the HER[2c] and increase the chemical stability of Co+ intermediate species,[4] which is a prerequisite for a large catalytic turnover number. Electromeric characteristic of halogen atoms resemble those of (per)fluoroaryl groups,[5] so that suitably decorated clathrochelates can show similar electrocatalytic activity but have the stability that is even higher.[6] Indeed, (per)fluorinated iron and cobalt(II) clathrochelates with pendant (per)fluoroarylsulfide substituents are known to withstand harsh acidic conditions.[7] In these complexes, however, the presence of bridging sulfur atoms does not allow the (per)fluoroaryl groups to stabilize the reduced (and probably electrocatalytically active) metal(I)-containing intermediate.[7] Removing these atoms may potentially open a new pathway to better (more efficient and more stable) clathrochelate-based electrocatalysts for the HER, which are functional analogs of macrocyclic BF2-cross-linked iron(II) complexes of perfluoroaryl-substituted α-dioximates earlier used for this purpose.[8] Here, we report iron and cobalt(II) (per)fluoroclathrochelates with inherent perfluoroaryl ribbed substituents, which can be obtained by a simple one-pot synthetic approach, and their behavior in the HER.

Experimental Section

Materials and Physical Measurements

The reagents used, n class="Chemical">Fen>Cl2·4H2O, CoCl2·6H2O, n-butylboronic, phenylboronic and pentafluorophenylboronic acids, sorbents, and organic solvents, were obtained commercially (SAF). 1,2-Bis(perfluorophenyl)ethane-1,2-dion dioxime (perfluoro-α-benzildioxime, denoted as (C6F5)2GmH2), was prepared from dichloroglyoxime as described in ref[8], using a modified synthetic procedure (see below).

Analytical data (C, H, and N contents) were obtained with a Cclass="Chemical">n>an class="Chemical">arlo Erba model 1106 microanalyzer. Iron, boron, and fluorine contents were determined spectrophotometrically. The cobalt content was determined by X-ray fluorescence analysis.

Matrix-asn class="Chemical">sin>sted laser desorption/n>an class="Disease">ionization time-of-flight (MALDI-TOF) mass spectra in positive and negative ranges were recorded with a MALDI-TOF-MS Bruker Autoflex II (Bruker Daltonics) mass spectrometer in a reflecto-mol mode. Ionization was induced by an UV-laser with the wavelength 337 nm. The samples were applied to a nickel plate; 2,5-dihydroxybenzoic acid (DHB) was used as a matrix. The accuracy of measurements was 0.1%.

IR spectra of the solid samn class="Chemical">ples (KBr tablets) in the range of 400–4000 cm–1 were recorded with a Nicolet Magna-IR 750 Fourier-transform infrared spectrophotometer.

UV–vis spectra of their solutions in dichloromethane and class="Chemical">n>an class="Chemical">tetrahydrofuran (THF) were recorded in the range of 230–900 nm with a Varian Cary 50 spectrophotometer. Individual Gaussian components of these spectra were calculated with a Fityk program.[9]

n class="Chemical">1Hn>, 13C, and 19F nuclear magnetic resonance (NMR) spectra were recorded from CD2Cl2 and CD3CN solutions with a Bruker Avance 400 and 600 spectrometers. Chemical shifts were referenced relative to the residual signals of these deuterated solvents (1H 5.32 and 1.93 ppm and 13C 53.40 and 1.30 ppm for CD2Cl2 and CD3CN, respectively); 19F NMR chemical shifts were referenced to the external CFCl3.

X-band electron paramagnetic resonance (Eclass="Chemical">n>an class="Chemical">PR) spectra for the cobalt(II) complexes were acquired on a Bruker ESP 300E spectrometer. Their glassy samples were obtained from 5 mM dichloromethane solutions at the temperature of liquid nitrogen (77.4 K) using a quartz finger dewar. The EPR spectra were registered with the following parameters: microwave frequency, 9.55 GHz; microwave power, 5 mW; modulation frequency, 100 kHz; modulation amplitude, 2 G; conversion time, 16 ms; time constant, 20 ms; and resolution, 1024 points. The EPR spectra simulation was performed using EasySpin.[10]

Cyclic voltammetry (CV) experiments were n class="Chemical">performed for the acetonitrile solutions with 0.1M n>an class="Chemical">(n-C4H9)4NClO4 as the supporting electrolyte, using a Metrohm Autolab PGSTAT128N potentiostat with a conventional one-compartment three-electrode cell (5 mL of solution). A glassy carbon electrode (MF-2012, BASi), which has been used as the working electrode, was thoroughly polished with alumina slurry, sonicated for 2 min, and rinsed before every measurement. A platinum wire counter electrode and a standard Ag/AgCl/NaClaq reference electrode (RE-5B, BASi) were used. To account for the drift of the reference electrode, ferrocene was used as an internal standard, and all the measured potentials are reported relative to the Fc/Fc+ redox couple. The solutions were thoroughly deaerated by passing argon through them before the CV experiments and above these solutions during the measurements.

Magnetic measurements were performed using a Quantum Design PPMS-9 device under an applied dc field of 1 kOe. Finely ground microcrystalline powders were immobilized in a mineral oil matrix inside a polyethylene capsule. The magnetic data were corrected for the sample holder, the mineral oil, and the diamagnetic contribution.

Gas chromatography analysis of the gases evolved during the electrolyclass="Chemical">n>an class="Chemical">sis was performed with a Chromatec-Crystal 5005.2 gas chromatograph equipped with a thermal conductivity detector. The hydrogen content in the gaseous mixture was quantitatively detected using a 0.5 m-in-length 60/80 Carboxen-1000 column with an internal diameter 3 mm at 200 °C for the detector and at 60 °C for the oven. The carrier gas was argon flowing at a rate of 15 mL min–1. The injections (250 μL) were performed via a sampling loop. The retention time of gaseous H2 was 1.61 min.

n class="Chemical">Con> K-edge XANES and X-ray absorption fine structure (EXAFS) spn>ectra for the n>an class="Chemical">cobalt(II) complex Co((C6F5)2Gm)3(BC6F5)2 (3) and for its reduced derivative ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7) were measured at the Structural Materials Science beamline of the Kurchatov Synchrotron Radiation Source (NRC “Kurchatov Institute”, Moscow).[11] White synchrotron beam was monochromatized with a Si(111) channel-cut monochromator. Beam intensities before and after the samples were measured with ion chambers filled with nitrogen–argon mixtures to provide 20 and 80% transmittance, respectively. The energy scale was calibrated against the experimental spectrum of Co foil measured under identical conditions by assigning the energy of 7709 eV to the maximum derivative point. Preliminary data processing and analysis, including nonlinear curve fitting with ab initio theoretical standards, were performed using Athena and Artemis codes from the IFEFFIT software package.[12]

Synthesis

(C6F5)2GmH2

n class="Chemical">Magnesiumn> turnings (4.0 g, 180 mmol) and THF (30 mL) were placed in a 500 mL flask, and a solution of C6F5Br (22.4 mL, 180 mmol) in THF (30 mL) was added dropwise to the boiling reaction mixture under intensive stirring. The reaction mixture was refluxed for 30 min, then cooled to −10 °C, and a solution of dichloroglyoxime (7.0 g, 45 mmol) in THF (25 mL) was added dropwise. The reaction mixture was stirred at this temperature for 1 h and then left overnight at r.t. The obtained dark-red solution was evaporated to approximately 30 mL and diluted with NH4Cl aqueous solution (10 g in 150 mL) under stirring. After evaporation of THF, the beige precipitate was filtered off and recrystallized from a methanol–water mixture to give a white fine-crystalline product. Yield: 8.1 g (68%). 1H NMR (CD3CN): δ (ppm) 10.41 (br s, 2H, NOH). 19F NMR (CD3CN): δ −105.13 (dt, 3JFF = 14.75, 20.17 Hz, 2F, m-F), −95.23 (t, 3JFF = 20.17 Hz, 1F, p-F), −81.62 (d, 3JFF = 14.75 Hz, 2F, o-F). 13C{1H} NMR (CD3CN): δ (ppm) 105.74 (td, 2JCF = 20.52 Hz, 3JCF = 3.59 Hz, i-C), 137.68 (dt, 1JCF = 250.02 Hz, 2JCF = 13.19 Hz, m-C), 142.03 (dm, 1JCF = 251.89 Hz, p-C), 142.42 (m, C=N), 143.27 (dm, 1JCF = 247.64 Hz, o-C). UV–vis (CH3OH) λmax, nm (ε × 10–3, mol–1 L cm–1): 221 (18), 240 (1.0), 247 (2.5), 264 (1.4), 312 (0.2), 337 (0.1).

Fe((C6F5)2Gm)3(Bn-C4H9)2 (1)

n class="Chemical">(C6F5)2GmH2n> (0.196 g, 0.47 mmol) and n-butylboronic acid (0.032 g, 0.32 mmol) were dissolved/suspended in nitromethane (20 mL), and FeCl2·4H2O (0.03 g, 0.15 mmol) was added to the stirring reaction mixture under argon. The refluxing reaction mixture was stirred for 5 h, evaporated to a small volume (approximately 2 mL), and precipitated with 5% HCl aqueous solution (20 mL) under cooling at 4 °C. The precipitate was filtered off, washed with 5% HCl aqueous solution (20 mL, in three portions), water (30 mL, in three portions), and ethanol (20 mL), and then extracted with dichloromethane (10 mL). The extract was flash-chromatographically separated on silica gel (30 mm layer; eluent: dichloromethane). The first red elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.156 g (78%). Spectroscopic and thin-layer chromatography characteristics of this product are identical to those reported in ref[6].

Fe((C6F5)2Gm)3(BC6H5)2 (2)

n class="Chemical">Fen>Cl2·4H2O (0.037 g, 0.19 mmol) and (C6F5)2GmH2 (0.25 g, 0.60 mmol) were dissolved/suspended in trifluoroacetic acid (10 mL), and phenylboronic acid (0.05 g, 0.41 mmol) was added to the reaction mixture under stirring in argon. The refluxing reaction mixture was stirred for 30 min and then evaporated to dryness. The solid residue was extracted with dichloromethane (40 mL), the extract was washed with water (60 mL, in six portions) and dried with CaCl2. The solution was evaporated to a small volume (approximately 4 mL) and flash-chromatographically separated on silica gel (30 mm layer, eluent: dichloromethane). The dichloromethane elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.065 g (26%). M = 1486.13. Anal. Calcd for C54H10N6O6B2F30Fe (%): C, 43.64; H, 0.68; N, 5.66; Fe, 3.76. Found (%): C, 43.81; H, 0.79; N, 5.57; Fe, 3.80. MS (MALDI-TOF) m/z: 1486 [M+]+•. 1H NMR (CD2Cl2): δ, ppm 7.24 (m, 8H, Ph), 7.30 (m, 2H, Ph). 13C{1H} NMR (CD2Cl2): δ, ppm 103.64 (t, 2JCF = 18 Hz, i-C6F5), 128.06 (s, m-Ph), 128.66 (s, i-Ph), 130.61 (s, p-Ph), 130.79 (s, o-Ph), 137.71 (d, 1JCF = 257 Hz, m-C6F5), 139.91 (s, C6F5C=N), 143.36 (d, 1JCF = 260 Hz, p-C6F5), 144.32 (d, 1JCF = 254 Hz, o-C6F5). 19F NMR (CD2Cl2): δ, ppm −76.9 (d, 2F, 2J = 19 Hz, ortho-C6F5), −88.7 (t, F, 2J = 19 Hz, para-C6F5), −100.7 (t, 2F, 2J = 20 Hz, meta-C6F5), −109.6 (m, 2F, O3BF). 11B NMR (CD2Cl2): δ, ppm 3.85 (d, 1J = 17 Hz). IR (KBr) ν (cm–1): 850, 906, 944, 997, 1061, 1112 ν(N–O) + ν(C–F), 1212 (m) ν(B–O), 1535 ν(C=N). UV–vis (CH2Cl2) λmax, nm (ε × 10–3, mol–1 L cm–1): 259 (47), 267 (0.9), 314 (3.4), 334 (3.5), 427 (0.6), 449 (9.2), 459 (13).

Co((C6F5)2Gm)3(BC6H5)2 (3)

n class="Chemical">Pn>henylboronic acid (0.06 g, 0.50 mmol) and (C6F5)2GmH2 (0.31 g, 0.73 mmol) were dissolved/suspended in nitromethane (10 mL), and CoCl2 (0.03 g, 0.24 mmol) was added to the reaction mixture under stirring in argon. The refluxing reaction mixture was stirred for 3 h and then evaporated to a small volume (approximately 4 mL) and precipitated with water (10 mL). The precipitate was filtered off, washed with water (15 mL, in four portions), ethanol (20 mL, in two portions) and hexane (10 mL), and then extracted with dichloromethane (10 mL). The extract was flash-chromatographically separated on silica gel (30 mm layer; eluent: dichloromethane) and once again separated in the same manner (70 mm layer of SiO2; eluent: dichloromethane–hexane 1:2 mixture). The second red elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.124 g (35%). M = 1489.21. Anal. Calcd for C64H10F30N6O6B2Co (%): C, 43.55; H, 0.68; N, 0.68. Found (%): C, 43.39; H, 0.60; N, 5.57. MS (MALDI-TOF) m/z (I, %): 1489 (100) [M]+•, 1637 (85) [M + DHB – H2O]+•, 1785 (15) [M + 2DHB – 2H2O]+•. 1H NMR (CD2Cl2): δ, ppm 6.27 (br s, 4H, o-H), 6.71 (br s, 4H, m-H), 6.98 (br s, 2H, p-H). 19F NMR (CD2Cl2): δ, ppm −99.15 (br s, 12F, m-F), −89.88 (br s, 6F, p-F), −80.38 (br s, 12H, o-F). 13C{1H} NMR (CD2Cl2): δ, ppm 98.42 (br s, o-C6F5), 116.26 (br s, i-C6H5), 128.08 (s, m-C6H5), 128.22 (s, p-C6H5), 131.69 (br s, o-C6H5), 140.74 (d, 1JCF = 259.98 Hz, m-C6F5), 145.34 (d, 1JCF = 252.91 Hz, p-C6F5). IR (KBr) ν (cm–1): 861, 906, 943, 996, 1113 ν(N–O) + ν(C–F), 1214 (m) ν(B–O), 1533 ν(C=N). UV–vis (CH2Cl2) λmax, nm (ε × 10–3, mol–1 L cm–1): 248 (56), 259 (6.2), 320 (6.1), 368 (8.4), 375 (1.2), 461 (4.4), 484 (5.7).

Co((C6F5)2Gm)3(Bn-C4H9)2 (4)

n class="Chemical">(C6F5)2GmH2n> (0.37 g, 0.88 mmol) and n-butylboronic acid (0.075 g, 0.73 mmol) were dissolved/suspended in nitromethane (10 mL), and CoCl2 (0.031 g, 0.24 mmol) was added to the reaction mixture under stirring in argon. The refluxing reaction mixture was stirred for 3 h and then evaporated to a small volume (of approximately 6 mL). The solution was diluted with ethanol (4 mL) and precipitated with water (10 mL) under cooling at 4 °C. The precipitate was filtered off, washed with water (20 mL, in three portions), and then extracted with dichloromethane (10 mL). The extract was flash-chromatographically separated on silica gel (30 mm layer; eluent: dichloromethane) and once again separated in the same manner (70 mm layer of SiO2; eluent: dichloromethane–hexane 1:10 mixture). The second dark-red elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.098 g (28%). M = 1449.23. Anal. Calcd for C60H18F32N6O6B2Co (%): C, 41.44; H, 1.25; N, 5.80. Found (%): C, 41.36; H, 1.33; N, 5.73. MS (MALDI-TOF) m/z: 1449 [M]+•. 1H NMR (CD2Cl2): δ, ppm −1.41 (br s, 4H, CH2B), 0.26 (br s, 4H, 2-CH2), 0.57 (br s, 4H, 3-CH2), 0.45 (br s, 6H, CH3). 19F NMR (CD2Cl2): δ, ppm −99.65 (br s, 12F, m-F), −90.71 (br s, 6F, p-F), −80.26 (br s, 12H, o-F). 13C{1H} NMR (CD2Cl2): δ, ppm −3.66 (br s, CH2B), 13.99 (s, CH3), 26.10 (s, 3-CH2), 27.08 (s, 2-CH2), 99.97 (br s, o-C), 140.94 (d, 1JCF = 260.3 Hz, m-C), 145.19 (d, 1JCF = 254.3 Hz, p-C). IR (KBr) ν (cm–1): 840, 908, 1001 ν(N–O), 1113 (m) ν(B–O) + ν(C–F), 1532 ν(C=N). UV–vis (CH2Cl2) λmax, nm (ε × 10–3, mol–1 L cm–1): 250 (46), 263 (7.4), 306 (0.9), 320 (3.1), 340 (3.9), 371 (4.0), 378 (4.6), 468 (4.5), 484 (4.7).

Fe((C6F5)2Gm)3(BC6F5)2 (5)

n class="Chemical">Pn>entafluorophenylboronic acid (0.136 g, 0.64 mmol) and (C6F5)2GmH2 (0.324 g, 0.77 mmol) were dissolved/suspended in nitromethane (15 mL), and FeCl2·4H2O (0.046 g, 0.23 mmol) was added to the reaction mixture under stirring in argon. The refluxing reaction mixture was stirred for 3 h and then evaporated to a small volume (of approximately 2 mL). The solution was diluted with ethanol (5 mL), precipitated with water (20 mL) under cooling at 4 °C, and filtered off. The precipitate was washed with a water–ethanol 5:1 mixture (20 mL, in three portions) and extracted with dichloromethane (10 mL). The extract was flash-chromatographically separated on silica gel (30 mm layer; eluent: dichloromethane), the second orange-red elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.194 g (55%). M = 1666.03. Anal. Calcd for C48F40N6O6B2Fe (%): C, 38.93; N, 5.04; Fe, 3.35. Found (%): C, 38.97; N, 5.17; Fe, 3.23. MS (MALDI-TOF) m/z: 1689 [M + Na+]+, 1705 [M + K+]+. 19F NMR (CD2Cl2): δ, ppm −103.94 (t, 3JFF = 21.6 Hz, m-F(ap)), −99.82 (t, 3JFF = 17.3 Hz, m-F(rib)), −94.71 (t, 3JFF = 19.5 Hz, p-F(ap)), −86.46 (t, 3JFF = 19.5 Hz, p-F(rib)), −76.82 (d, 3JFF = 18.1 Hz, o-F(rib)), −75.86 (t, 3JFF = 16.7 Hz, o-F(ap)). 13C{1H} NMR (CD2Cl2): δ, ppm 102.65 (t, 2JCF = 16.4 Hz, i-C(rib)), 137.25 (d, 1JCF = 248.2 Hz, o-C(ap)), 137.70 (d, 1JCF = 256.6 Hz, m-C(rib)), 141.65 (d, 1JCF = 248.2 Hz, p-C(ap)), 142.34 (s, C=N), 143.70 (d, 1JCF = 262.1 Hz, p-C(rib)), 144.31 (d, 1JCF = 256.6 Hz, o-C(rib)), 148.82 (d, 1JCF = 247.2 Hz, m-C(ap)). IR (KBr) ν (cm–1): 853, 907, 997, 1064, 1112 ν(N–O) + ν(C–F), 1158 (m) ν(B–O), 1535 ν(C=N). UV–vis (CH2Cl2) λmax, nm (ε × 10–3, mol–1 L cm–1): 244 (60), 252 (9.7), 338 (1.5), 364 (2.1), 447 (11), 459 (12).

Co((C6F5)2Gm)3(BC6F5)2 (6)

n class="Chemical">Pn>entafluorophenylboronic acid (0.136 g, 0.64 mmol) and (C6F5)2GmH2 (0.324 g, 0.77 mmol) were dissolved/suspended in nitromethane (20 mL), and CoCl2 (0.029 g, 0.22 mmol) was added to the reaction mixture under stirring in argon. The refluxing reaction mixture was stirred for 3 h and then evaporated to a small volume (approximately 2 mL). The solution was diluted with ethanol (5 mL), precipitated with water (20 mL) under cooling at 4 °C, and filtered off. The precipitate was washed with a water–ethanol 5:1 mixture (20 mL, in three portions) and extracted with dichloromethane (10 mL). The extract was flash-chromatographically separated on silica gel (30 mm layer; eluent: dichloromethane) and once again separated in the same manner (70 mm layer of SiO2; eluent: dichloromethane–hexane 2:3 mixture). The second elute was evaporated to dryness, and the solid residue was dried in vacuo. Yield: 0.108 g (30%). M = 1669.12. Anal. Calcd for C54F40N6O6B2Co (%): C, 38.86; N, 5.04; F, 45.53; Co, 3.53. Found (%): C, 38.69; N, 4.90; F, 45.34; Co, 3.42. MS (MALDI-TOF) m/z: 1669 [M]+•. 19F NMR (CD2Cl2): δ, ppm −80.18 (d, 12F, 3JFF = 21.4 Hz, o-F(rib)), −79.09 (d, 4F, 3JFF = 21.4 Hz, o-F(ap)), −89.33 (t, 6F, 3JFF = 20.5 Hz, p-F(rib)), −96.59 (t, 2F, 3JFF = 19.9 Hz, p-F(ap)), −99.15 (m, 12F, m-F(rib)), −105.63 (m, 4F, m-F(ap)). 13C{1H} NMR (CD2Cl2): δ, ppm 100.6 (br s, o-C(rib)), 140.54 (d, 1JCF = 258.0 Hz, m-C(rib)), 144.32 (d, 1JCF = 242.6 Hz, p-C(rib)), 138.08 (d, 1JCF = 250.0 Hz, p-C(ap)), 136.05 (d, 1JCF = 250.0 Hz, m-C(ap)), 148.34 (d, 1JCF = 240.1 Hz, o-C(ap)). IR (KBr) ν (cm–1): 872, 911, 960, 1000, 1114 ν(N–O) + ν(C–F), 1140 (m) ν(B–O), 1534 ν(C=N). UV–vis (CH2Cl2) λmax, nm (ε × 10–3, mol–1 L cm–1): 236 (52), 254 (7.4), 260 (1.0), 362 (6.1), 374 (3.6), 465 (4.4), 482 (5.2).

((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7)

n class="Chemical">Con>mplex 3 (0.25 g, 0.17 mmol), silver powder (0.09 g, 0.83 mmol), and (CH3)4NCl (0.02 g, 0.18 mmol) were dissolved/suspended in dry acetonitrile (15 mL) under argon. The dark-blue reaction mixture was stirred overnight and then evaporated to dryness. The solid residue was extracted with toluene (30 mL), the extract was filtered under argon and then evaporated to dryness. The solid residue was dried in vacuo. Yield: 0.16 g (61%). Anal. Calcd for C58H22N7F30O6B2Co (%): C, 44.56; H, 1.42; N, 6.27. Found (%): C, 44.46; H, 1.58; N, 6.44. MS (MALDI-TOF, negative range) m/z: −1489 [M – (CH3)4N+]−. 1H NMR (CD3CN): δ, ppm 6.12 (br s, o-H), 6.46 (br s, m-H), 6.62 (br s, p-H). 19F NMR (CD3CN): δ, ppm −113.61 (br s, o-F), −102.03 (br s, p-F), −98.67 (br s, m-F). 13C NMR (CD3CN): δ, ppm 122.06 (d, 1JCF = 252.7 Hz, ar-C(rib)), 144.36 (d, 1JCF = 264.2 Hz, ar-C(rib)), 165.46 (d, 1JCF = 241.2 Hz, m-C(rib)). UV–vis (THF) λmax, nm (ε × 10–3, mol–1 L cm–1): 251 (52), 291 (3.7), 315 (7.3), 364 (5.5), 385 (1.1), 490 (2.2), 575 (3.0), 661 (6.6).

Chemical formulas for the n class="Chemical">con>mplexes synthen>an class="Chemical">sized, the numbers they are referred to in the following discussion, electronic configurations, and spin states are collected in Table .

Table 1

A List of Cage Complexes under Study Detailing Their Electronic Configurations and Spin States

compound	given number	electronic configuration	spin
Fe((C6F5)2Gm)3(Bn-C4H9)2	1	3d6	0
Fe((C6F5)2Gm)3(BC6H5)2	2	3d6	0
Co((C6F5)2Gm)3(BC6H5)2	3	3d7	1/2
Co((C6F5)2Gm)3(Bn-C4H9)2	4	3d7	1/2
Fe((C6F5)2Gm)3(BC6F5)2	5	3d6	0
Co((C6F5)2Gm)3(BC6F5)2	6	3d7	1/2
((CH3)4N)[Co((C6F5)2Gm)3(BC6H5)2]	7	3d8	1
Fe(I2Gm)3(Bn-C4H9)2	8	3d6	0

X-ray Crystallography

n class="Chemical">Sin>ngle crystals of the complexes Co((C6F5)2Gm)3(BC6H5)2·CH2Cl2 (3·CH2Cl2), Co((C6F5)2Gm)3(BC6F5)2·CH2Cl2 (6·CH2Cl2), and Co((C6F5)2Gm)3(Bn-C4H9)2 (4) were grown at room temperature from their solutions in dichloromethane–hexane and benzene–iso-octane mixtures, respectively. The intensities of reflections were measured at 120.0(2) K with a Bruker Apex II charge-coupled device diffractometer using Mo Kα (for 4, λ = 0.71073 Å) and Cu Kα (for 3 and 6, λ = 1.54178 Å) radiation. The structures were solved by the direct method and refined by full-matrix least squares against F2. Nonhydrogen atoms were refined in anisotropic approximation. Hydrogen atoms were included in the refinement by the riding model with Uiso(H) = nUeq(C), where n = 1.5 for methyl groups and 1.2 for the other atoms. The unit cell of the complex 6·CH2Cl2 contains four solvate dichloromethane molecules, which have been treated as a diffuse contribution to the overall scattering without specific atom positions by SQUEEZE/PLATON.[13] All calculations were made using the SHELXTL[14] and OLEX2[15] program packages. The crystallographic data and experimental details are listed in Table S1 (see Supporting Information). CCDC 1061775–1061777 contain the supplementary crystallographic data.

Results and Discussion

The first cage n class="Chemical">con>mplex of this series, n>an class="Chemical">Fe((C6F5)2Gm)3(Bn-C4H9)2 (1), has been synthesized[6] in a two-step procedure that includes perfluoroarylation with Cu(C6F5) of the hexaiodoclathrochelate precursor Fe(I2Gm)3(Bn-C4H9)2 (8, I2Gm2– is diiodoglyoxime dianion), which needs to be isolated for each type of apical capping groups before that. We failed to isolate its cobalt(II) analogues because of their lower thermodynamic stability and side redox reactions they undergo under vigorous reaction conditions.

At the same time, pentafluoron class="Chemical">phenylboron-cappn>ed iron and cobalt(II) hexachloroclathrochelates Fe(Cl2Gm)3(BC6F5)2 and Co(Cl2Gm)3(BC6F5)2 (Cl2Gm2– is dichloroglyoxime dianion) have been recently synthesized[7] by a one-pot template condensation of a weakly donor dichloroglyoxime with pentafluorophenylboronic acid with the metal ion as a matrix and trifluoroacetic acid as a solvent (under vigorous reaction conditions). The purpose of using this rather unusual solvent was to increase the activity of C6F5B(OH)2 as a capping (cross-linking) Lewis-acidic agent and to prevent it from undergoing a deborylation reaction. Besides, the n-butylboron-capped analogues of such complexes have been earlier described[16] to form under regular reaction conditions.

In the present study, n-butyl-, phenyl-, and pentafluorophenylboron-capn>ped n>an class="Chemical">iron and cobalt(II) tris-perfluoro-α-benzildioximates were synthesized by Scheme using the template condensation of three molecules of α-dioxime with the corresponding boronic acid on the metal(II) ion (Fe2+ or Co2+) as a matrix. The reaction was performed under vigorous reaction conditions (with boiling nitromethane or trifluoroacetic acid used as the solvent and with distillation of a solvent–water azeotrope), and the target iron and cobalt(II) clathrochelates were isolated in moderate yields (25–55%).

Scheme 1

Synthesis of Iron and Cobalt (Per)fluoroclathrochelates and Their Precursors

The CV data (see below) suggested the stability of the n class="Gene">Co(I)n> intermediates in the CV time scale, we attempted to obtain a chemically reduced n>an class="Chemical">cobalt(I) clathrochelate using the synthetic approach described earlier.[4] The reduction of Co((C6F5)2Gm)3(BC6H5)2 (3) by an excess of powder silver in the presence of tetramethylammonium chloride (in combination with which the metallic silver is known to form a strong reducing system[16]) in acetonitrile led to the formation of a dark-blue solution with the intensive coloration caused by the clathrochelate anion [Co((C6F5)2Gm)3(BC6H5)2]−. This anion was isolated as a salt with the bulky tetramethylammonium cation, a navy blue solid product ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7), that was air-stable for several months but rapidly oxidized in a solution.

All the clathrochelates obtained were then chn class="Chemical">arn>acterized by elemental analysis, MALDI-TOF mass spectrometry, IR, UV–vis, CV, EPR, 1H, 13C, and 19F NMR spectroscopies, and single crystal X-ray diffraction (for 3·CH2Cl2, 4, and 6·CH2Cl2).

UV–vis spectra of the obtained tris-pn>erfluoro-α-benzildioximate iron and cobalt(II) clathrochelates contain, in their visible range, two intensive bands assigned to a metal-to-ligand Md → Lπ* charge transfer (MLCT). The bands in the spectra of the iron(II) complexes are significantly shifted (by approximately 25 nm) in the UV range and are more intensive (ε ≈ 1 ÷ 1.5 × 104 mol–1 L cm–1) than the corresponding MLCT bands in the cobalt(II) complexes (ε ≈ 5 × 103 mol–1 L cm–1). At the same time, they all are shortwave-shifted (by approximately 20 nm) as compared to those of their α-benzildioximate (i.e. nonperfluorinated) analogues (Table S2 in the Supporting Information). The bands of π–π* intraligand transitions in the UV–vis range for two types of the cage complexes (i.e. the macrobicyclic metal(II) perfluoro- and α-benzildioximates) are also shifted relative to each other, thus showing a significant redistribution of the electron density in the quasiaromatic cage framework as a result of the perfluorinated ribbed substituents.

Reduction of the encapsulated cobalt(II) ion is rclass="Chemical">n>an class="Gene">esponsible for the change in the color of the tris-perfluoro-α-benzildioximate complexes from dark brown to navy blue (Figure ). The same blue color, which has also been observed for the cobalt(I) clathrochelates with nitrogen-containing ligands,[4,16] stems from two highly intensive bands that appear at approximately 660 (ε = 6.6 × 103 mol–1 L cm–1) and 575 nm (ε = 3.0 × 103 mol–1 L cm–1) (Figure ) and correspond to the metal-to-ligand Cod → Lπ* backdonation and the ligand-to-metal Lπ → Cod charge transfer. Upon reduction of the encapsulated cobalt(II) ion to cobalt(I), these charge transfer bands shift to the longwave region by 115–175 nm; by contrast, the intraligand π–π* transition bands in the UV region are slightly shortwave-shifted.

Figure 1

UV–vis spectra of THF solutions of the parent cobalt(II) clathrochelate Co((C6F5)2Gm)3(BC6H5)2 (3, in brown) and its reduced cobalt(I) derivative ((CH3)4N)+[CoI((C6F5)2Gm)3(BC6H5)2]− (7, in blue) at the same concentrations.

UV–vis spectra of n class="Chemical">pan class="Chemical">THF solutions of the pn>arent cobalt(II) clathrochelate Co((C6F5)2Gm)3(BC6H5)2 (3, in brown) and its reduced cobalt(I) derivative ((CH3)4N)+[CoI((C6F5)2Gm)3(BC6H5)2]− (7, in blue) at the same concentrations.

IR spectra of all the obtained iron and class="Chemical">n>an class="Chemical">cobalt(II) (per)fluoroclathrochelates contain the N–O, B–O, and C=N stretching vibration bands, which are characteristic of the boron-capped tris-dioximate clathrochelates, and those of the C–F bonds characteristic of their pentafluoroaryl-ribbed substituents.

In the MALDI-TOF spectra of the cobalt(II) clathrochelate intraclass="Chemical">n>an class="Chemical">complexes, the peak of the molecular ion always dominates over those of its adducts with the DHB matrix in their positive ranges. The spectrum of a cobalt(I)-encapsulating ionic associate ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7) contains in its negative range (Figure S6) an intensive peak of the clathrochelate anion at −1489 amu.

The number and position of the class="Chemical">n>an class="Chemical">signals in 1H, 19F{1H}, and 13C{1H} NMR spectra in solution, together with the ratios of their integral intensities, also confirmed the composition of the obtained iron and cobalt(II) clathrochelates. Chemical shifts of the nuclei in the apical substituents are similar to those in analogous butyl-,[17] phenyl-,[16] and perfluorophenyl-containing clathrochelates[7] with other ribbed substituents. On the other hand, the chemical shifts of the nuclei in the ribbed substituents are similar between the clathrochelates with other apical substituents. Therefore, apical and ribbed substituents only slightly affect the electronic environment of the nuclei of each other.

NMR sn class="Chemical">pectra of the obtained cobalt(II) perfluorophenylclathrochelates n>an class="Chemical">are typical for low-spin cobalt(II) cage complexes.[17] Paramagnetic shifts of their nuclei compared to those in the diamagnetic iron(II)-encapsulating analogues are relatively small (below 2 ppm for protons and less than 4 ppm for 19F and 13C nuclei); the only exception is the ortho-carbon of the ribbed substituents. The latter is close to the encapsulated paramagnetic metallocenter, the cobalt(II) ion, so it has a paramagnetic shift of 45 ppm owing to the significant direct spin density delocalization to its nucleus. Note that the signals of the ipso-carbon nuclei in 13C NMR spectra could not be observed because of the paramagnetic broadening. In addition, a dynamic Jahn–Teller exchange between three possible distorted molecular structures of each complex at room temperature leads to partial averaging of both the Fermi and dipolar contributions to the paramagnetic shifts.[17]

X-band En class="Chemical">Pn>R spectrosn>an class="Chemical">copy confirmed the low-spin nature of the obtained cobalt(II) clathrochelates; their spectra (Figure ) contain a well-resolved eight-line splitting in the downfield region caused by the hyperfine interaction with the 59Co nucleus (I = 7/2). Both g and hyperfine tensors are rhombic, and their values are characteristic of the low-spin cobalt(II) clathrochelates.[17] Note that high-spin cobalt(II) clathrochelates are EPR-silent in the X-band as a result of a very large negative zero-field splitting.[18]

Figure 2

Experimental and simulated EPR spectra of 1 mM dichloromethane solution of the cobalt(II) clathrochelate Co((C6F5)2Gm)3(BC6F5)2 (6, X-band, 80 K). The parameters of simulation are as follows: g = 1.99, g = 2.10, g = 2.26, CoA = 145 MHz, CoA = 15 MHz, and CoA = 420 MHz.

Experimental and simulated Eclass="Chemical">n>an class="Chemical">PR spectra of 1 mM dichloromethane solution of the cobalt(II) clathrochelate Co((C6F5)2Gm)3(BC6F5)2 (6, X-band, 80 K). The parameters of simulation are as follows: g = 1.99, g = 2.10, g = 2.26, CoA = 145 MHz, CoA = 15 MHz, and CoA = 420 MHz.

No EPR class="Chemical">n>an class="Chemical">signal was observed for the isolated cobalt(I) perfluorophenyl clathrochelate ((CH3)4N)+[CoI((C6F5)2Gm)3(BC6H5)2]− (7) in the X-band at 4 and 78 K, as typical for high-spin cobalt(I) clathrochelates with large positive zero-field splitting,[4,16] which makes them EPR-silent. The high-spin nature of Co(I) ion in this complex is also consistent with the data from variable-temperature dc magnetic susceptibility measurements (Figure ): at 300 K, χMT is 1.16 cm3 K mol–1, which is only slightly larger than the spin-only value for s = 1.

Figure 3

Variable temperature magnetic susceptibility data for a microcrystalline sample of ((CH3)4N)+[CoI((C6F5)2Gm)3(BC6H5)2]− (7) collected under an applied dc field of 1 kOe.

Vn class="Chemical">arn>iable temperature magnetic susceptibility data for a microcrystalline sample of ((CH3)4N)+[CoI((C6F5)2Gm)3(BC6H5)2]− (7) collected under an applied dc field of 1 kOe.

The paramagnetic nature of [class="Chemical">n>an class="Gene">CoI((C6F5)2Gm)3(BC6H5)2]− species also follows from the NMR spectroscopy. Its NMR spectra 1H, 19F{1H}, and 13C{1H} are dominated by the paramagnetic shifts, which are larger than in the above cobalt(II) clathrochelates, as there is no dynamic Jahn–Teller distortion in the cobalt(I) perfluorophenyl clathrochelate but it has a nonzero pseudocontact contribution arising from its large zero-field splitting. The alternation in the direction of the paramagnetic shifts, which is observed for the nuclei of the ribbed fragments, also suggests the significant contact contribution.

Note that the molecular structures of the clathrochelates class="Chemical">n>an class="Chemical">Co((C6F5)2Gm)3(BC6H5)2 (3), Co((C6F5)2Gm)3(Bn-C4H9)2 (4), and Co((C6F5)2Gm)3(BC6F5)2 (6) were additionally confirmed by single crystal X-ray diffraction (Figures –6, Table S3). According to its results, the Co–N distances in these fluorinated cobalt(II) clathrochelates vary by 0.21 Å because of the Jahn–Teller distortion, so that the metal ion is significantly shifted from the center of the “cage” to one of the −N=C–C=N– ribbed fragments. The CoN6-coordination polyhedron in Co((C6F5)2Gm)3(BC6F5)2 (6) is close to the trigonal prism (TP, Scheme ; the distortion angle φ = 1.4°), whereas in the other two (φ = 10°–14°), it adopts a geometry that is intermediate between a TP (φ = 0°) and a trigonal antiprism (TAP, φ = 60°). For comparison, the φ value in the iron(II) complex Fe((C6F5)2Gm)3(Bn-C4H9)2 (1) is equal to 25.4°,[6] and the metal ion is almost in the center of a cage framework. The degree of this TAP–TP distortion is affected not only by the nature of the metal ion but also by the nature of the ribbed substituents. Thus, the coordination polyhedra in the nonperfluorinated cobalt(II) tris-α-benzildioximates with the same apical groups (Bn-C4H9[17] and BC6H5[2a]) are closer to TAP (φ = 13.1° and 16.0°, respectively); however, the φ value in the corresponding iron(II) n-butylboron-capped clathrochelate varies only a little if the phenyl group is used instead of its ribbed perfluorophenyl (24.6°[17]). At the same time, perfluorination makes the aryl ribbed substituents to rotate relative to the α-dioximate fragments: average angles between their mean planes change from 55.7° and 58.3° in Co((C6F5)2Gm)3(Bn-C4H9)2 (4) and Co((C6F5)2Gm)3(BC6H5)2 (3), respectively, to 45.8°[17] and 46.7°[2a] in their α-benzildioximate analogs. In Fe((C6F5)2Gm)3(Bn-C4H9)2 (1), the same angle is 58.9°,[17] which changes to 42.6° upon going to the corresponding α-benzildioximate iron(II) complex.[17] As a result of this rotation, the mutual mesomeric effects of the perfluoroaryl substituents and the quasiaromatic polyazomethine cage framework cancel out, so that no π-conjugation between them is observed.

Figure 4

General view of Co(C6F5Gm)3(Bn-C4H9)2 (4, a; hereinafter, nonhydrogen atoms are shown as thermal ellipsoids at p = 50%) and its CoN6-coordination polyhedron (b) with Co–N distances (Å).

Figure 6

General view of Co((C6F5)2Gm)3(BC6F5)2 (6, a) and its CoN6-coordination polyhedron (b) with Co–N distances (Å).

Scheme 2

TP–TAP Distortion of a MN6-Coordination Polyhedron

General view of n class="Chemical">Co(C6F5Gm)3(Bn-C4H9)2n> (4, a; hereinafter, nonhydrogen atoms are shown as thermal ellipsoids at p = 50%) and its CoN6-coordination polyhedron (b) with Co–N distances (Å).

General view of pan class="Chemical">Co((C6F5)2Gm)3(BC6H5)2n> (3, a) and its pan class="Chemical">CoN6-pan class="Chemical">coordination polyhedron (b) with Co–N distances (Å).

General view of pan class="Chemical">Co((C6F5)2Gm)3(BC6F5)2n> (6, a) and its pan class="Chemical">CoN6-pan class="Chemical">coordination polyhedron (b) with Co–N distances (Å).

In the absence of X-ray diffraction data for the n class="Gene">Co(I)n> complex (as many attempts to grow its single crystals failed miserably), additional confirmation for its formation comes from a comparative study of CoII((C6F5)2Gm)3(BC6H5)2 (3) and ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7) by X-ray absorption spectroscopy (Figure ). The spectrum for the CoI complex is clearly shifted to a lower energy with respect to its CoII counterpart: the energy positions of the absorption maxima are 7727.3 and 7729.0 eV, respectively. This shift is especially apparent in the derivative spectra (shown as an inset in Figure ), which feature three distinct maxima at 7707.2, 7716.7, and 7721.9 eV for ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7) and at 7707.2, 7717.8, and 7723.7 eV for CoII((C6F5)2Gm)3(BC6H5)2 (3). Such a shift is consistent with the formal reduction and the resultant decrease in the local electrostatic potential at the cobalt nuclei in the CoI compound.

Figure 7

Co K-edge spectra of CoII((C6F5)2Gm)3(BC6H5)2 (3, in black) and ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7, in red); the inset shows the first derivatives of the absorption coefficient.

n class="Chemical">Con> K-edge spectra of n>an class="Gene">CoII((C6F5)2Gm)3(BC6H5)2 (3, in black) and ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7, in red); the inset shows the first derivatives of the absorption coefficient.

FTs of the n class="Chemical">Con> K-edge spectra for the same two n>an class="Chemical">compounds are shown in Figure . There is a significant difference in the intensity and shape of the first FT peak corresponding to the Co–N coordination sphere (Figure ). In the case of ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7), this peak is higher and shifted to longer distances as compared to CoII((C6F5)2Gm)3(BC6H5)2 (3). To achieve a good fit for the latter compound, it was necessary to assume the existence of four short (1.89 Å) and two long (2.07 Å) Co–N bonds, that is, a 4 + 2 coordination in a fair agreement with the direct X-ray diffraction data (see Table S3), demonstrating a strongly Jahn–Teller distorted TP environment of the cobalt(II) ion. In the case of ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7), a good fit is obtained with a single Co–N distance of 1.97 Å and a coordination number of 6.

Figure 8

Amplitudes of FTs of Co K-edge EXAFS spectra for CoII((C6F5)2Gm)3(BC6H5)2 (3, in black) and ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7, in red); the solid lines correspond to the experimental data and the open circles denote the best-fit theoretical curves.

Amplitudes of FTs of Co K-edge EXAFS spclass="Chemical">n>ectra for CoII((C6F5)2Gm)3(BC6H5)2 (3, in black) and ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7, in red); the solid lines correspond to the experimental data and the open circles denote the best-fit theoretical curves.

Electrochemical pron class="Chemical">perties of the obtained iron and n>an class="Chemical">cobalt(II) (per)fluoroclathrochelates were studied using CV. The corresponding CVs contain a single cathodic wave in the potential range from 0 to −1 V versus the Fc/Fc+ couple, which is assigned to the Co2+/+ reduction. In all cases, this wave is reversible (as follows from ΔEp = Ea – Ec being in the range of 60–70 mV and from the current ratio for the direct reduction and the reverse backward reoxidation processes equal to one) and shows a diffusional control, as its peak current depends linearly on the square root of the scan rate (Figure ). Therefore, the anionic cobalt(I)-containing species resulting from this metal-centered Co2+/+ reduction are stable on the CV time scale. The reduction potential slightly depends on the nature of the apical substituent in the cobalt clathrochelates. An increase in its electron-withdrawing effect going from n-butyl to phenyl and to perfluorophenyl shifts the potential to the anodic region from −0.415 to −0.314 and to −0.210 V (relative to the Fc/Fc+ redox couple), respectively. Note that for the corresponding iron clathrochelates, the reduction potentials are equal to −0.837, −0.749, and −0.691 V.

Figure 9

CVs (a) and the plots of the reduction and oxidation peak currents vs the square root of the scan rate (b) for 1 mM acetonitrile solution of the clathrochelate Co((C6F5)2Gm)3(Bn-C4H9)2 (4). Scan rates are from 0.1 to 1 V s–1 (T = 298 K), 0.1 M (n-(C4H9)4N)ClO4 as the supporting electrolyte. All the potentials are referenced to the Fc/Fc+ couple.

CVs (a) and the plots of the reduction and oxidation n class="Chemical">peak currents vs the square root of the scan rate (b) for 1 mM n>an class="Chemical">acetonitrile solution of the clathrochelate Co((C6F5)2Gm)3(Bn-C4H9)2 (4). Scan rates are from 0.1 to 1 V s–1 (T = 298 K), 0.1 M (n-(C4H9)4N)ClO4 as the supporting electrolyte. All the potentials are referenced to the Fc/Fc+ couple.

The electrocatalytic activity in the HER was tested for all the obtained n class="Chemical">ironn> and cobalt(II) clathrochelates using different organic and inorganic acids (including acetic, trifluoroacetic, trifluoromethanesulfonic, and perchloric acids) as a source of H+ ions. In contrast to previously described clathrochelate-based electrocatalysts,[2] the addition of acid to the acetonitrile solutions of these complexes did not produce any electrocatalytic enhancement of the current. Moreover, the addition of up to 5 equiv of acetic or trifluoroacetic acids did not alter the CV response of the system; full reversibility of the redox event remained even at scan rates as low as 20 mV s–1 (Figure S1). In the presence of strong trifluoromethanesulfonic (Figure S2) and perchloric (Figure S3) acids, however, the reduction became irreversible, again without any significant enhancement in the reduction current. In the latter case, the addition of more than 2 equiv of the acid caused an additional positively shifted oxidation peak to appear on the reverse CV scan, suggesting the instability of the metal(I) complexes in the solutions of very strong acids rather than any electrocatalytic behavior. Indeed, further addition of these strong acids results in the destruction of the original metal(II) complexes, as judging by the loss of the color of their solutions. No electrocatalytic enhancement of the current was observed after the acid-induced decomposition of the metal(II) complexes. Even if electrocatalytically active metal nanoparticles were formed in the process,[3] they were also not stable in these harsh acidic conditions. The bulk 30 min electrolysis performed for 1 mM acetonitrile solutions of cobalt(II) clathrochelates in the presence of 5 equiv of trifluoroacetic and trifluoromethanesulfonic acids showed no evolution of the molecular hydrogen detected by gas chromatography. Note that in the case of stronger trifluoromethanesulfonic acid, the solutions of cobalt(II) clathrochelates lose their color in the very beginning of the experiment, suggesting that under these harsh acidic conditions both electrochemically generated cobalt(I) and parent cobalt(II) complexes are unstable.

To get inpan class="Chemical">sin>ght into the stability of the pan class="Chemical">cobalt(I) pan class="Chemical">complex in less acidic solutions, we employed NMR spectroscopy. The addition of up to 3 equiv of trifluoroacetic acid to the acetonitrile-d3 solution of ((CH3)4N)[CoI((C6F5)2Gm)3(BC6H5)2] (7) under anaerobic conditions did not result in noticeable changes in both 1H and 19F NMR spectra even after 12 h. Further increase in the concentration of the acid to 5 equiv led to the fast transformation of the cobalt(I) complex into the parent cobalt(II) clathrochelate (Figures S4 and S5).

No signal of moleculclass="Chemical">n>an class="Chemical">ar hydrogen was detected in the 1H NMR spectra, suggesting the reduction of some other substrate. Although a detailed investigation into the mechanism of the reaction between the cobalt(I) complex and trifluoroacetic acid is out of scope of the present study (as being irrelevant to the HER), the 1H NMR data hint on the possible reduction of the solvent under these conditions: the signal at 6.0 ppm (Figure S5b) is characteristic of the NH4+ cation (the splitting to three equal lines with the coupling constant of 53 Hz is due to the interaction with the quadrupolar 14N nuclei), which may be produced either from the initial clathrochelate or as a result of the reduction of acetonitrile. While conducting the experiment in a nitrogen-free dioxane-d8 solution, no such signal was observed and the kinetics of CoI/CoII transition was much slower (full conversion in 8 h), thus suggesting the latter mechanism.

Thus, dn class="Gene">espn>ite the effective stabilization of the metal(I)-encapsulating reduced species by the perfluorophenyl ribbed substituents, the corresponding iron and cobalt(II) clathrochelates are inactive in the electrocatalytic HER. If the electrocatalysis is assumed to occur via the homogenous mechanism,[2] a possible reason for this may be an almost complete isolation of the encapsulated metal ion from the environment by bulky hydrophobic C6F5 substituents, which do not allow it to be protonated. At higher concentrations of the acid, however, the protonation seems to take place and to produce a dead-end complex, which is catalytically inactive in the HER (although, apparently, on a longer time scale, it converts back to the initial cobalt(II) compound with the reduction of some other substrate). On the other hand, a very high stability of the cobalt(I)-containing clathrochelate species may prevent their destruction as long as the acidic conditions are not extremely harsh, so that the electrocatalytically active metal nanoparticles[3] can no longer form from them.

Conclusions

Although the perfluoron class="Chemical">phenyl ribbed substituents in the iron and n>an class="Chemical">cobalt(II) clathrochelates with different apical groups help stabilizing the metal(I)-encapsulating reduced species, these complexes turned out to be not electrocatalytically active in the HER. Such an unexpected behavior may stem either from the encapsulated metallocenter being well-isolated from the environment by the bulky hydrophobic perfluorinated substituents or from the intermediate metal(I) species being effectively stabilized by the caging ligands, which would make both the homogenous and heterogenous electrocatalysis less probable. However, the electrochemical reversibility of the reduction of all the complexes obtained offers new opportunities for the isolation of other chemically stable metal(I) clathrochelates; these studies are currently underway in our group.